【摘要】 利用遥感数据,综合最大似然法监督分类、多尺度空间分层聚类的部分监督分类方法和主成分方法,分析黄河上游龙羊峡水库库区1987～1999年间土地利用土地覆盖变化。提取专题信息,不同要素采用不同方法;具体分类中,土地利用类型的一级类型耕地、水体及未利用土地类型采用主成分分析和最大似然法监督分类方法;对一级类型草地采用多尺度分层聚类算法的部分监督分类方法。结果表明,草地信息利用SSHC方法提取结果较好,与Bayes分类方法相比,精度提高4.2%,SSHC所获结果数据Kappa系数为0.84,Bayes所获结果数据Kappa系数为0.78。对某专题要素分类,此方法结果较优。更多还原

【Abstract】 The Longyangxia Reservoir is the largest hydraulic power and water conservancy project in the up stream of the Yellow River, with integrated functions of generating electricity, irrigating farmland and controling flood. People pay great attention to the land use/land cover(LUCC) of region around the reservoir, the sand quantity entering the reservoir, land desertification, grass land degradation and so on. The multi-temporal and multi-spectral remote sensed data were used in this study. According to the eco-environment characteristics of the study area as well as finished study work, integrated with SSHC, MLS and PCA, the land-use and land-cover change of Longyangxia Reservoir area was analyzed from 1987 to 1999. The different methods were adopted to extract different thematic information of environmental factors. The MLS and PCA methods were applied in obtaining information of cropland, water area and other non-use lands. The supervise classification based on SSHC was used to gain grassland information. The result showed that the effect for extracting grassland information using SSHC is well, the accuracy of classification may increase 4.2% compared with Bayes′ classification, the Kappa coefficient of SSHC is 0.84, while the Kappa coefficient of Bayes is 0.78. Thus it can be seen that SSHC is well than other methods in some aspects in imagery classification.更多还原